What Is Value at Risk?
VaR compresses the risk of an entire portfolio into a single dollar figure defined by a horizon and a confidence level. A bank might report a one-day 99 percent VaR of $25 million, meaning that under normal market conditions it expects to lose more than $25 million on roughly one trading day out of a hundred.
The measure became the common language of market risk in the 1990s after J.P. Morgan popularized its RiskMetrics methodology, and regulators later embedded VaR-style measures into bank capital rules. Its appeal is comparability: a risk manager can aggregate equities, bonds, currencies, and derivatives into one number and track it daily against limits.
How VaR Is Calculated
The parametric approach assumes returns follow a normal distribution and computes VaR from the portfolio's volatility; at 95 percent confidence, VaR is roughly 1.65 standard deviations of the portfolio's value. Historical simulation instead replays actual past market moves, often the last 250 to 500 trading days, and reads the loss at the chosen percentile. Monte Carlo simulation generates thousands of random scenarios from modeled distributions and does the same.
Each method trades off simplicity against realism. Parametric VaR is fast but understates fat-tailed risks, historical simulation captures real crises but only ones that already happened, and Monte Carlo is flexible but depends entirely on modeling assumptions. Desks typically scale one-day VaR to longer horizons by multiplying by the square root of time.
Why It Matters and Its Limits
VaR sets the boundaries of risk-taking across Wall Street. Trading desks operate under VaR limits, hedge funds report VaR to investors, and banks hold regulatory capital tied to stressed versions of the measure. Anyone interviewing for sales and trading or risk management should be able to define VaR precisely and interpret a figure like a one-day 99 percent VaR of $5 million.
The metric's critical flaw is that it says nothing about how bad losses get beyond the threshold, which is why the 2008 crisis discredited naive reliance on it. Complements such as expected shortfall, which averages losses in the tail, and stress testing against scenarios like 2008 or the 2020 crash are now standard alongside VaR.
